U.S. patent application number 14/674742 was filed with the patent office on 2016-03-24 for generation of support data records using natural language processing.
The applicant listed for this patent is BMC SOFTWARE, INC.. Invention is credited to Sun Chun CHU, Troy CLINE, Jonathan W. HALL, Nilesh PHADKE.
Application Number | 20160085856 14/674742 |
Document ID | / |
Family ID | 54238219 |
Filed Date | 2016-03-24 |
United States Patent
Application |
20160085856 |
Kind Code |
A1 |
HALL; Jonathan W. ; et
al. |
March 24, 2016 |
GENERATION OF SUPPORT DATA RECORDS USING NATURAL LANGUAGE
PROCESSING
Abstract
A view generator receives support text characterizing a support
requirement for available information technology (IT) support, the
support text being received in sentence form via a graphical user
interface (GUI). A text analyzer performs natural language
processing on the support text and thereby identifies at least one
sentence part and at least one named entity within the support
text. A support record generator relates each of the at least one
sentence part and the at least one named entity to a support record
type, and generates a support data record for the support
requirement, including filling individual fields of the support
data record using the at least one sentence part and the at least
one named entity.
Inventors: |
HALL; Jonathan W.; (Reading,
GB) ; CHU; Sun Chun; (Richmond Hill, CA) ;
CLINE; Troy; (Leander, TX) ; PHADKE; Nilesh;
(Pune, IN) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
BMC SOFTWARE, INC. |
Houston |
TX |
US |
|
|
Family ID: |
54238219 |
Appl. No.: |
14/674742 |
Filed: |
March 31, 2015 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
62053496 |
Sep 22, 2014 |
|
|
|
Current U.S.
Class: |
706/46 ;
707/739 |
Current CPC
Class: |
G06F 40/295 20200101;
G06F 16/3344 20190101; G06F 16/345 20190101; G06Q 10/06315
20130101; G06F 40/174 20200101; G06F 40/30 20200101; G06F 40/216
20200101; G06Q 10/20 20130101; G06N 7/005 20130101; G06F 16/353
20190101; G06F 16/3346 20190101 |
International
Class: |
G06F 17/30 20060101
G06F017/30; G06F 17/24 20060101 G06F017/24; G06N 7/00 20060101
G06N007/00; G06F 17/27 20060101 G06F017/27 |
Claims
1. A system including instructions stored on a non-transitory
computer readable storage medium and executable by at least one
processor, the system comprising: a view generator configured to
cause the at least one processor to receive support text
characterizing a support requirement for available information
technology (IT) support, the support text being received in
sentence form via a graphical user interface (GUI); a text analyzer
configured to cause the at least one processor to perform natural
language processing on the support text and thereby identify at
least one sentence part and at least one named entity within the
support text; and a support record generator configured to cause
the at least one processor to relate each of the at least one
sentence part and the at least one named entity to a support record
type, and further configured to cause the at least one processor to
generate a support data record for the support requirement,
including filling individual fields of the support data record
using the at least one sentence part and the at least one named
entity.
2. The system of claim 1, wherein the view generator is further
configured to provide the GUI including a single field dialog box
for receiving the support text.
3. The system of claim 1, wherein the text analyzer includes a
sentence parser configured to identify individual sentences within
the support text and parse the individual sentences to obtain
individual words and numbers.
4. The system of claim 1, wherein the text analyzer includes a
sentence part identifier configured to identify and label the at
least one sentence part within the support text, based on a context
of a use thereof within the support text.
5. The system of claim 1, wherein the text analyzer includes a
named entity extractor configured to execute at least one query
against support-related data to identify the at least one named
entity of the at least one sentence part.
6. The system of claim 1, wherein the support record generator is
configured to present the support record type among a plurality of
potential support record types, and to receive confirmation of
correct association of the support record type with the support
text, using the GUI.
7. The system of claim 1, wherein the support record generator is
configured to execute at least one machine learning algorithm to
classify the at least one sentence part and the at least one named
entity with respect to a plurality of possible support record
types, including the support record type.
8. The system of claim 1, wherein the support record generator
includes: a sentence part score calculator configured to provide a
sentence part score for the at least one sentence part, the
sentence part score representing a probabilistic determination that
the at least one sentence part is associated with each of a first
plurality of support record types, including the support record
type; and a named entity score calculator configured to provide a
named entity score for the at least one named entity, the named
entity score representing a probabilistic determination that the at
least one named entity is associated with each of a second
plurality of support record types, including the support record
type.
9. The system of claim 8, wherein the support record generator
includes a support record type identifier configured to combine the
sentence part score and the named entity score to identify the
support record type.
10. The system of claim 1, wherein the support record generator is
configured to generate an initial support record type, and includes
a score modifier configured to receive a modification of the
initial support record type, in response to a determination that
the initial support record type does not correspond to the support
text, wherein the score modifier is configured to adjust one or
both of the sentence part score and the named entity score and
thereby enable generation of the support record type based
thereon.
11. The system of claim 1, wherein the support record generator
includes a support record field mapper configured to fill
individual fields of the support record using the at least one
sentence part and the at least one named entity, including mapping
a label identifying the at least one sentence part and a name of
the at least one named entity. to corresponding ones of the
individual fields of the support data record.
12. A method comprising: receiving support text characterizing a
support requirement for available information technology (IT)
support, the support text being received in sentence form via a
graphical user interface (GUI); identifying at least one sentence
part within the support text; calculating a sentence part score
relating the at least one sentence part to at least on support type
of the available IT support; identifying at least one named entity
within the support text, including executing a query against
support-related data, using the support text; calculating a named
entity score relating the at least one named entity to the at least
one support type; and filling the at least one sentence part and
the at least one named entity within corresponding fields of a
support data record of the at least on support type.
13. The method of claim 12, wherein the identifying at least one
sentence part includes performing natural language processing of
the support text.
14. The method of claim 12, wherein the executing at least one
query against support-related data to identify the at least one
named entity of the at least one sentence part.
15. A computer program product including instructions recorded on a
non-transitory computer readable storage medium and configured,
when executed by at least one semiconductor processor, to cause the
at least one semiconductor processor to: receive support text
characterizing a support requirement for available information
technology (IT) support, the support text being received in
sentence form via a graphical user interface (GUI); identify at
least one sentence part within the support text; calculate a
sentence part score relating the at least one sentence part to at
least on support type of the available IT support; identify at
least one named entity within the support text, including executing
a query against support-related data, using the support text;
calculate a named entity score relating the at least one named
entity to the at least one support type; and fill the at least one
sentence part and the at least one named entity within
corresponding fields of a support data record of the at least on
support type.
16. The computer program product of claim 15, wherein the
instructions, when executed, are further configured to cause the at
least one processor to: execute at least one machine learning
algorithm to classify the at least one sentence part and the at
least one named entity with respect to a plurality of possible
support record types, including the support record type.
17. The computer program product of claim 15, wherein the
instructions, when executed, are further configured to cause the at
least one processor to: fill individual fields of the support data
record using the at least one sentence part and the at least one
named entity, including mapping a label identifying the at least
one sentence part and a name of the at least one named entity, to
corresponding ones of the individual fields of the support data
record.
18. The computer program product of claim 15, wherein the
instructions, when executed, are further configured to cause the at
least one processor to: present the support record type among a
plurality of potential support record types, and to receive
confirmation of correct association of the support record type with
the support text, using the GUI.
19. The computer program product of claim 15, wherein the
instructions, when executed, are further configured to cause the at
least one processor to: provide a sentence part score for the at
least one sentence part, the sentence part score representing a
probabilistic determination that the at least one sentence part is
associated with each of a first plurality of support record types,
including the support type; and provide a named entity score for
the at least one named entity, the named entity score representing
a probabilistic determination that the at least one named entity is
associated with each of a second plurality of support record types,
including the support type.
20. The computer program product of claim 19, wherein the
instructions, when executed, are further configured to cause the at
least one processor to: combine the sentence part score and the
named entity score to identify the support type.
Description
CROSS REFERENCE TO RELATED APPLICATIONS
[0001] This application claims the benefit of U.S. Provisional
Application No. 62/053,496, filed on Sep. 22, 2014, and titled
"Generation of Support Data Records Using Natural Language
Processing", which is hereby incorporated by reference in its
entirety.
TECHNICAL FIELD
[0002] This description relates to the providing of Information
Technology (IT) support.
BACKGROUND
[0003] Information technology (IT) generally includes any scenario
in which computers and related technologies are deployed to
facilitate a productivity, convenience, or enjoyment of a user or a
group of users. In many cases, however, such users do not possess
sufficient technical skills to fully realize associated benefits.
Moreover, even when users have required technical skills, the use
of computer hardware and software deployed within an IT landscape
may be subject to organizational rules and other constraints
imposed by a deploying organization.
[0004] As a result, users may not be capable of utilizing IT
resources in a desired or required fashion. Thus, IT support
professionals are often tasked with providing service and other
support to such users. For example, users may require assistance
with malfunctioning hardware or software, or may be unable to
complete a required upgrade or other modification with respect to
such hardware or software.
[0005] In a typical workflow for providing IT support, an IT
support professional generally interacts with the user in question,
and works to determine a type of support that may be required,
along with associated details that may be useful in providing the
required support. Further, the IT support professional may create a
data record that captures all such gathered information in a
standardized format. By creating such data records, it is
relatively straightforward, for example, to track a progress over
time toward resolution for the user, to compare the
currently-required support with similar, previous support
instances, and to facilitate potential involvement of a second IT
support professional (if needed).
SUMMARY
[0006] According to one general aspect, a system includes
instructions stored on a non-transitory computer readable storage
medium and executable by at least one processor. The system
includes a view generator configured to cause the at least one
processor to receive support text characterizing a support
requirement for available information technology (IT) support, the
support text being received in sentence form via a graphical user
interface (GUI). The system includes a text analyzer configured to
cause the at least one processor to perform natural language
processing on the support text and thereby identify at least one
sentence part and at least one named entity within the support
text. The system includes a support record generator configured to
cause the at least one processor to relate each of the at least one
sentence part and the at least one named entity to a support record
type, and further configured to cause the at least one processor to
generate a support data record for the support requirement,
including filling individual fields of the support data record
using the at least one sentence part and the at least one named
entity.
[0007] According to another general aspect, a method includes
receiving support text characterizing a support requirement for
available information technology (IT) support, the support text
being received in sentence form via a graphical user interface
(GUI). The method includes identifying at least one sentence part
within the support text, calculating a sentence part score relating
the at least one sentence part to at least on support type of the
available IT support, and identifying at least one named entity
within the support text, including executing a query against
support-related data, using the support text. The method further
includes calculating a named entity score relating the at least one
named entity to the at least one support type, and filling the at
least one sentence part and the at least one named entity within
corresponding fields of a support data record of the at least on
support type.
[0008] According to another general aspect, a computer program
product includes instructions recorded on a non-transitory computer
readable storage medium and configured, when executed by at least
one semiconductor processor, to cause the at least one
semiconductor processor to receive support text characterizing a
support requirement for available information technology (IT)
support, the support text being received in sentence form via a
graphical user interface (GUI). The instructions, when executed,
further cause the at least one semiconductor processor to identify
at least one sentence part within the support text, and calculate a
sentence part score relating the at least one sentence part to at
least on support type of the available IT support. The
instructions, when executed, further cause the at least one
semiconductor processor to identify at least one named entity
within the support text, including executing a query against
support-related data, using the support text, calculate a named
entity score relating the at least one named entity to the at least
one support type, and fill the at least one sentence part and the
at least one named entity within corresponding fields of a support
data record of the at least on support type.
[0009] The details of one or more implementations are set forth in
the accompanying drawings and the description below. Other features
will be apparent from the description and drawings, and from the
claims.
BRIEF DESCRIPTION OF THE DRAWINGS
[0010] FIG. 1 is a block diagram of a system for providing IT
support.
[0011] FIG. 2 is a flowchart illustrating example operations of the
system of FIG. 1.
[0012] FIG. 3 is a block diagram of an example implementation of
portions of the system of FIG. 1.
[0013] FIG. 4 is a flowchart illustrating more detailed example
operations of the systems of FIGS. 1 and 3.
[0014] FIG. 5 is a table illustrating example scoring operations
for sentence parts in the system of FIG. 1.
[0015] FIG. 6 is a table illustrating example scoring operations
for named entities in the system of FIG. 1.
[0016] FIG. 7 is a first table illustrating scoring modification
used in the system of FIG. 1.
[0017] FIG. 8 is a second table illustrating further example
scoring modifications used in the systems of FIG. 1 and FIG. 7.
[0018] FIG. 9 is a table illustrating example techniques for
presenting and confirming identified support record types.
[0019] FIG. 10 is a table illustrating example techniques for
populating data fields of support records.
DETAILED DESCRIPTION
[0020] FIG. 1 is a block diagram of a system 100 for providing IT
support. In the example of FIG. 1, a support server 102 facilitates
the providing of IT support to one or more users, in conjunction
with a support client 104. More particularly, as described in
detail below, the support server 102 enables fast, convenient, and
accurate description of an IT task to be performed. Moreover, the
support server 102 may be configured to transform such a
description into one or more support data records that may be used,
as referenced above, to facilitate prompt and complete resolution
of the IT support task in question.
[0021] In operation, the support server 102 and the support client
104 may operate either locally or remotely with respect to one
another. In example embodiments, an IT support professional may
utilize the support client 104 while interacting with a user (e.g.,
customer, consumer, or employee) who requires support, and/or while
utilizing a problem description received from such a user.
[0022] For example, as shown, the support server 102 may include a
view generator 106 that is configured to provide a support dialog
box 108 in conjunction with a graphical user interface (GUI)
associated with the support client 104. In the example of FIG. 1,
the support dialog box 108 is illustrated as including support text
110, which represents text providing the problem description
referenced above. More specifically, as described in detail below,
the support dialog box 108 may be implemented as a single dialog
box configured to capture all required text of an associated
problem description. In other implementations, however, some of
which are described below, various other techniques may be utilized
to capture text associated with the problem description.
[0023] In further detail, as shown, the support server 102 may
include a text analyzer 112 that is configured to analyze the
support text 110, to thereby identify portions thereof that are
determined to be potentially instrumental in enabling a support
record generator 114 to generate a plurality of support records
116. In this regard, as referenced above and described in more
detail below, the support records 116 generally represent
individual data records stored in their appropriate table or other
database or repository, where such support records each correspond
with at least one of the various types of support data records that
may be relevant to the practice of providing user support within an
IT landscape.
[0024] Thus, by way of non-limiting example, the support records
116 may include general customer requests, customer support
tickets, and other records defined with respect to established IT
service management standards. For example, such IT service
management standards may include incident tickets, problem records,
known errors, requests for change (RFCs), and work orders. For
example, a required format for the support records 116 may be
implemented based on common framework standards, such as, for
example, the IT infrastructure library (ITIL).
[0025] In practice, various techniques and scenarios for providing
IT support are well known. To provide a few non-limiting examples,
it may occur that the IT support professional, also referred to
herein as a support provider, is engaged in current interaction
with one or more users. For example, the support provider may
interact with a user by way of telephone, email, or chat
functionality. Thus, such communication may be synchronous or
a-synchronous. For example, in some scenarios, the IT support
provider may actively elicit the support text 110 from a user, and
may therefore enter the support text 110 into the support dialog
box 108. In other implementations, the IT support provider may
receive some or all of the support text 110 by way of an electronic
transmission received from the user. In some implementations, the
IT support provider and the user may be co-located, perhaps
utilizing a single computing device. In other cases, the support
provider may utilize a local computing device to remotely access or
control the computing device of the user. In the latter scenarios,
the user device being controlled is typically the device requiring
IT support. In other implementations, however, the user may or may
not be utilizing the device requiring support.
[0026] In these and many other potential scenarios, some of which
are referenced below, the support server 102 utilizes the support
text 118 to automatically create a corresponding support record of
the support records 116. As described, such a support record may
then be used as a basis for managing or resolving delivery of a
particular service or other IT support function, such as a
resolution of an IT issue, or a managed action and review of a
requested or required change.
[0027] More particularly, the support server 102, e.g., the text
analyzer 112 and the support record generator 114, may be
configured to determine a required record type, from among a
selection that is available and relevant to the IT environment in
which the support provider is working. For example, in an IT
service context, the support server 102 may determine whether a
relevant record type is one related to resolution of a service
outage (also referred to as an incident), or a request to perform a
piece of routine service activity that is not related to an
unexpected issue, often referred to as a work order.
[0028] The support server 102 may further be configured to assist
the IT support provider in gathering of certain types of support
information from the user, without which the resulting support
record may be incomplete. Additionally, or alternatively, the IT
support server 102 may collect such information directly from the
user, independently of the support provider. Types of information
referenced here may include, e.g., common discrete pieces of
information, such as an identity of the user, or specific
identifiers of any related assets or other items (e.g., a unique
identification or name of a specific computer or other piece of
hardware, in the context of identifying a technical issue related
thereto).
[0029] In practice, such information may vary widely, in accordance
with a type of IT support issue and associated record type.
Moreover, particularly when users are permitted to enter the
support text 110 directly, the support text 110 may include an
overly or needlessly verbose description of the IT task to be
performed and/or issue to be resolved. Nonetheless, as described in
detail herein, the support server 102 may be configured to obtain
and utilize the support text 110 in a manner that facilitates fast,
accurate, complete, and partially or completely automatic
generation of the support records 116.
[0030] In order to provide these and other features, as already
referenced, the support server 102 may include a view generator 106
that provides the support dialog box 108. In specific
implementations, the support dialog box 108 may represent a single
field for entry of the support text 110 at the support client 104.
In other words, the support client 104 does not require, but may
utilize if desired, any other type of entry mechanism for obtaining
the support text 110, including, for example, dropdown menus,
multiple discrete/defined text entry fields, or any other known of
future technique for capturing the support text 110 from the
user.
[0031] Advantageously, in such implementations in which only the
single support dialog box 108 is utilized, the user and/or the
support provider may experience a reduction or prevention of a need
to actively switch between fields or screens of the support client
104. As a result, for example, the support text 110 may be received
in a smooth and natural manner, without interruption caused by any
requirement to move between different screen items, or different
screens.
[0032] Once the support text 110 has been received, or as the
support text 110 is being received in real-time or near real-time,
the text analyzer 112 may proceed to analyze the support text 110,
e.g., using particular naturalized language processing techniques.
To facilitate and enable these and various other advantages, the
text analyzer 112 may be configured to interpret the support text
110, using language analysis, such as natural language processing
(NLP). For example, a sentence parser 118 of the text analyzer 112
may be configured to analyze individual sentences of the support
text 110, and identify individual words, punctuation marks,
numbers, or any other included symbols, while removing spaces,
paragraph breaks, or any other unnecessary portions of the support
text 110.
[0033] Using the output of the sentence parser 118, a sentence part
identifier 120 may identify individual words, phrases,
abbreviations, numbers, or other identifiable sentence parts. For
example, as described in more detail below, the sentence part
identifier 120 may compare individual output from the sentence
parser 118 with a table of pre-identified sentence parts, to
thereby label at least some of the individual outputs of the
sentence parser 118 in accordance with the table. For example, a
word identified by the sentence parser 118 may be mapped to such a
table for identification of the word as a verb, noun, or other part
of speech, or as a name. Similarly, a number may be mapped to
identify the number as an IP address, or as a model number for a
particular type of hardware. Of course, additional or alternative
techniques may be used to implement the sentence part identifier
120, some of which are described in more detail below.
[0034] Meanwhile, a named entity extractor 122 may be configured to
receive outputs of the sentence parser 118 and/or the sentence part
identifier 120. As shown, a query generator 124 of the named entity
extractor 122 may be configured to generate query results 126 by
executing one or more queries against various types of
support-related data 128. More specifically, and as also described
in more detail below, the named entity extractor 122 is operable to
identify, in a fast and accurate fashion, a potentially large
number of entities named within the support-related data 128, even
in scenarios in which some or all of the support-related data 128
is known to change over time.
[0035] For example, as referenced above, the sentence parser 118
may identify a word that the sentence part identifier 120
identifies as a name. Then, the query generator 124 may execute a
query against the support-related data 128, whereupon the query
results 126 may indicate that the name belongs to a customer
identified as such within the support-related data 128. Of course,
in the example, the named entity extractor 122 could also execute
the just-described search based directly on an output of the
sentence parser 118, as well.
[0036] Although the support-related data 128 is illustrated in the
simplified example of FIG. 1 as being a single element within the
support server 102, it should be appreciated that the
support-related data 128 generally represents any potential,
relevant data source of a potential plurality of data sources that
may contain data useful to the named entity extractor 122, and,
moreover, any or all such data sources may be located remotely from
the support server 102. For example, the support-related data 128
may include virtually any organizational data owned or accessible
by an organization deploying the support server 102. In other
examples, the support-related data 128 could also include virtually
any publicly-available information that may be useful to the named
entity extractor 122 in identifying named entities of interest.
[0037] The support record generator 114 may thus be configured to
receive the identified sentence parts from the sentence part
identifier 120, as well as the named entities from the named entity
extractor 122, to thereby generate the support records 116 (or, at
least, to provide an initial version thereof that is sufficient to
facilitate and enable ongoing IT support efforts for one or more
support requirements associated with the support text 110). In
other words, in at least some implementations, the support record
generator 114 is generally configured to receive identified
sentence parts and named entities, and to classify the received
sentence parts and named entities with respect to specific types of
support records 116, and/or with respect to specific fields of
individual support records.
[0038] To facilitate these and related operations, in the example
of FIG. 1, a trainer 130 is illustrated in conjunction with a
training set 132, and may be utilized to implement one or more
machine-learning algorithms, to thereby enable the just-referenced
classification functions of the support record generator 114. For
example, as referenced in more detail below, the training set 132
may generally represent a plurality of support records, such as a
subset of the support records 116, and/or any other data that may
be considered to be useful or representative with respect to
classifying portions of the support text 110.
[0039] Then, the trainer 130 may be configured, perhaps in
conjunction with human assistance, to include labels within the
training set 132 and in conjunction with data included therein, to
thereby enable the types of classification procedures referenced
above and described in detail below with respect to the support
record generator 114. Such training and associated classification
techniques may be implemented using a variety of known or future
techniques, some of which are referenced in more detail, below.
[0040] For example, such techniques may include the use of one or
more machine learning and/or data mining algorithms. By way of
non-limiting example, such algorithms may include, e.g., support
vector machines (SVM), Bayesian networks, neural networks,
regression-based networks, decision tree learning algorithms, or
combinations thereof. In some such scenarios, or related scenarios,
the training set 132 may include existing support records that have
already been labeled in a manner to facilitate operations of the
support record generator 114 (e.g., for supervised machine
learning), while, in other implementations, the training set 132
may include initially-unlabeled support records, which may be
utilized by the trainer 132 to deduce relevant data structures
therein (e.g., for unsupervised learning).
[0041] Thus, in the example of FIG. 1, the support record generator
114 includes a sentence part score calculator 134 and a named
entity score calculator 136. As the names imply, the sentence part
score calculator 134 may be configured to calculate a score for
each sentence part provided by the sentence part identifier 120,
while the named entity score calculator 136 may similarly be
configured to provide a score for each named entity provided by the
named entity extractor 122. That is, as described in detail below,
each such calculated score may generally indicate a relative
likelihood that a corresponding sentence part or named entity
should be associated with a particular type of support record, or
individual field thereof.
[0042] Consequently, a support record type identifier 138 may be
configured to receive calculated scores from the score calculators
134, 136, and calculate therewith an aggregated likelihood that the
support text 110 is related to one or more specific types of
support records, or individual fields thereof. For example, in a
highly simplified scenario, the sentence part score calculator 134
may indicate that a particular sentence part identified by the
sentence part identifier 120 is 70% likely to be associated with a
"work order" support record, and 30% likely to be associated with a
"request for change" support record. Meanwhile, the named entity
score calculator 136 may indicate that a named entity provided by
the named entity extractor 122 is 80% likely to be associated with
a "work order" support record, and 20% is likely to be associated
with a "service request" support record. Then, in the aggregate
simplified example, the support record type identifier 138 may
determine that the support text 110 from which the relevant
sentence part and named entity were determined may be highly likely
to be associated with a "work order" support record.
[0043] Thus, the sentence part score calculator 134 provides a
sentence part score for the sentence parts, the sentence part score
representing a probabilistic determination that a given sentence
part is associated with each of a first plurality of support record
types. Similarly, the named entity score calculator 136 provides a
named entity score for the named entities, the named entity score
representing a probabilistic determination that each named entity
is associated with each of a second plurality of support record
types. That is, although not explicitly shown in the examples
provided herein, the support record types identified by the
sentence part score calculator 134 need not align exactly with
those identified by the named entity score calculator 136.
[0044] As may be appreciated from the above description of the
trainer 130 and the training set 132, all of the sentence parts
score calculator 134, the named entity score calculator 136, and
the support record type identifier 138 may be implemented using one
or more trained classifiers implementing the types of
machine-learning or data mining algorithms referenced above. That
is, for example, in some cases, individual algorithms or other
techniques may be selected for implementation by individual ones of
the sentence part score calculator 134, the named entity score
calculator 136, and the support record type identifier 138.
Further, in some implementations, more simplistic techniques may be
used. For example, the support record type identifier 138 may be
executed using weighted or other parameterized aggregations of the
various scores provided by the score calculators 134, 136, without
requiring separate training by the trainer 130.
[0045] In some implementations, as referenced above and described
in more detail below, the support record generator 114 may
facilitate human involvement of varying types and extent during
generation of the support records 116. For example, a score
modifier 140 may be configured to enable a user of the system 100
to make adjustments to some or all of the scores provided by the
score calculators 134, 136 and/or the record types identified by
the support record type identifier 138. Specific examples of such
score adjustments are provided below, e.g., with respect to FIGS. 4
and 7-9.
[0046] Finally with respect to the support record generator 114, a
support record field mapper 142 may be configured to map specific
portions of the support text 110 into individual data fields of
instances of support records of the support record type identified
by the support record type identifier 138, and in conjunction with
scores calculated by the score calculators 134, 136. That is, the
support record field mapper 142 may be configured to map a label
identifying a sentence part and a name of a named entity to
corresponding, individual fields of the support record. Thus, in
some implementations, the resulting support records 116 should be
virtually indistinguishable from support records of the same
support type generating using conventional techniques. In other
words, support records provided by the support record field mapper
142 should be entirely compatible with existing IT support
infrastructure.
[0047] In operation, the support record field mapper 142 may be
configured to produce entire, completed support records for
inclusion with the support records 116. In other implementations,
however, and depending on content of the original support text 110,
the support record field mapper 142 may generate one or more
initial support records having one or more empty or incorrect data
fields included therein. In such cases, the support record field
mapper 142 may be configured to interact, e.g., with either an
administrator or other user of the support dialog box 108, or of
the system 100 in general, in order to complete one or more
specific support records for inclusion thereof within the support
records 116.
[0048] In the example of FIG. 1, the support server 102 is
illustrated as a single server that includes at least one processor
102A and non-transitory computer readable storage medium 102B. As
may be appreciated, the support server 102 may of course represent
two or more support servers in communication with one another.
Similarly, the at least one processor 102A should be understood to
represent two or more processors operation in parallel, any one or
more of which may be configured to execute instructions stored on
the non-transitory computer readable storage medium 102B in order
to obtain, e.g., the view generator 106, the text analyzer 112, and
the support record generator 114.
[0049] Somewhat similarly, the various modules of the support
server 102 illustrated in the example of FIG. 1 are illustrated as
separate, discrete modules in communication with one another, as
needed. In various implementations, however, each such individual
module may be implemented as two or more sub-modules in
communication with one another, or, conversely, any two or more
such modules may be implemented as a combined, single module.
[0050] FIG. 2 is a flowchart 200 illustrating example operations of
the system 100 of FIG. 1. In the example of FIG. 2, operations
202-212 are illustrated as separate, sequential operations.
However, it may be appreciated that, in alternative
implementations, one or more additional or alternative operations
may be included, while one or more such operations may be omitted
and any or all such implementations of the flowchart 200, as
needed. Further, in all such implementations, it will be
appreciated that any two or more of the operations 202-212 may be
executed in a different order than that shown in FIG. 2, or in a
nested, iterative, looped, or branched fashion, or in partially or
completely overlapping or parallel manner.
[0051] In the example of FIG. 2, support text characterizing a
support requirement for available information technology (IT)
support may be received, this support text being received in
sentence form via a graphical user interface (GUI) (202). For
example, in FIG. 1, the view generator 106 may be configured to
receive the support text 110 by providing the support dialog box
108 within the support client 104. The support requirement may
refer to any generalized or specific action requested of an IT
support provider, many of which are referenced herein (e.g.,
request for change, work order, incident resolution, or upgrade
request). Specific examples by which the support text 110 may be
received through operations of the view generator 106 are provided
below, e.g., with respect to FIG. 3.
[0052] At least one sentence part within the support text may be
identified (204). For example, the sentence part identifier 120,
perhaps based on parsed sentence portions received from the
sentence parser 118, may identify the types of sentence parts
referenced above and described in detail below, e.g., with respect
to FIGS. 3-7.
[0053] A sentence part score relating the at least one sentence
part to at least one support type of the available IT support may
be calculated (206). For example, the sentence part score
calculator 134 may be configured to calculate a relative likelihood
that an identified sentence part is associated with a specific
support record type associated with the support records 116.
[0054] At least one named entity within the support text may be
identified, including executing a query against support-related
data, using the support text (208). For example, in some
implementations, the various sentence parts provided above may be
used to generate queries at the query generator 124, which may then
be applied against the various types of support-related data 128,
to thereby obtain the query results 126. As referenced above and
described in detail below with respect to FIGS. 4 and 5, a
resulting sentence part score may be matched to a particular
support type, using a trained classification module provided by the
trainer 130, or, alternatively, by performing a mapping of the
identified sentence parts to corresponding sentence part scores
that may have been pre-calculated and stored by the support server
102.
[0055] At least one named entity within the support text may be
identified, including executing a query against support-related
data, using the support text (208). For example, the named entity
extractor 122 may utilize sentence parts provided by the sentence
part identifier 120, or individual words, phrases, or other content
provided directly by the sentence parser 118 but not identified as
a sentence part in the sense described above with respect to the
sentence part identifier 120. Then, as also described, the query
generator 124 may execute a query for individual ones of potential
named entities against the various types of support-related data
128. In this way, the query results 126 may be obtained as
including individual named entities within the support text 110 and
previously characterized as such, directly or indirectly, within
the support-related data 128.
[0056] A named entity score relating the at least one named entity
to the at least one support type may be calculated (210). For
example, the named entity score calculator 136 may be configured to
determine a relative likelihood that a given named entity should be
associated with one or more types of support records.
[0057] The at least one sentence part and the at least one named
entity may be filled within corresponding fields of a support data
record of the at least one support type (212). For example, the
support record type identifier 138 may be configured to identify a
most-likely support type for the original support types 110 and
extracted, identified, and scored sentence parts/named entities,
for use of the associated scores in determining one or more
potential record support types that may be identified by the
support record type identifier 138. In this way, as also described
herein, the support record field mapper 142 may then proceed with
mapping individual words, sentence parts, and/or named entities
within individual support records of the support type identified by
the support record type identifier 138.
[0058] FIG. 3 is a block diagram of a system 300 illustrating
aspects of example implementations of the system 100 of FIG. 1. In
the example of FIG. 3, implementation aspects are illustrated for
scenarios in which a support server 302, representing a support
server 102, interacts remotely with a support client 304,
representing the support client 104 of FIG. 1.
[0059] FIG. 3 provides an example in which the text analyzer 112 of
the system 100 is implemented in conjunction with the support
dialog box 108 in an interactive and convenient manner. Further,
the implementation of FIG. 3 enables quick, easy, and accurate
corrections or other modifications to initial results provided by
the text analyzer 112. As a result, operations of the support
record generator 114, not specifically illustrated in the specific
example of FIG. 3, may subsequently be optimized or otherwise
facilitated, to thereby improve an accuracy and completeness of the
support records 116.
[0060] In the example, as shown, the support server 302 implements
a representational state transfer (REST) application program
interface (API) 306, which refers to an API constructed in the REST
architectural style for designing and implementing network
applications. More specifically, such REST-based techniques utilize
a stateless, client-server, cashable communications protocol, such
as the hypertext transfer protocol (HTTP), to design and implement
network applications. Such a REST architecture may utilize HTTP to
create, read, update, and/or delete data, and generally provide
full featured yet lightweight techniques for interacting with the
support client 304.
[0061] Of course, the REST API 306 of FIG. 3 is merely a
non-limiting example of implementations of the system 100. In other
example implementations, other techniques for executing network
applications may be used, such as remote procedure calls (RPC), or
other web service protocols (e.g., the simple object access
protocol (SOAP)).
[0062] Further in the example of FIG. 3, a natural language
processing (NLP) service 308 is illustrated as being executed in
conjunction with the REST API 306. More specifically, the NLP
service 308 should be understood to represent a parameterized NLP
service configured to implement some or all of the functionalities
described above with respect to the text analyzer 112 of FIG.
1.
[0063] In this context, natural language processing refers
generally to known processing techniques for inputting text or
other language expressions that are expressed in an informal or
conversational manner, and which are designed to account for the
many various ambiguities that may be associated with words, groups
of words, or numbers in such context. For example, such ambiguities
may include scenarios in which words are used as different parts of
speech (e.g., noun or verb), or otherwise have different meanings,
and must be judged based on a context of their occurrence or use.
In the example of FIG. 3, as may be appreciated from the above
discussion of the text analyzer 112 of FIG. 1, as well as from the
discussion below provided with respect to FIGS. 4-10, the NLP
service 308 may be parameterized for use in the IT support
scenarios of the system 100 of FIG. 1. For example, as referenced,
numbers of a particular format may be recognized as IP addresses,
and names may be recognized as customers who may potentially
require IT support, among other examples.
[0064] The REST API 306 should be understood to correspond
generally to the view generator 106 of FIG. 1, and to provide many
of the functionalities described with respect thereto. In
particular, for example, the REST API 306 may interact with a
browser 310 at the support client 304, utilizing a jQuery component
312. More specifically, the jQuery component 312 should be
understood to represent a java script component for executing an
implementation of the support dialog box 108 of FIG. 1, and thereby
provide for receipt of the support text 110.
[0065] More specifically, the jQuery component 312 may represent a
specific library of java script code designed to facilitate and
otherwise optimize various scripting functions associated with the
use of java script in the context of enabling client-side interface
interaction. Such java script libraries may be utilized, for
example, to enable and facilitate interactions using the
Asynchronous Javas Script plus XML (AJAX) techniques. Such AJAX
techniques refer to known, client-side techniques for enabling
asynchronous web applications used to interact with a server,
without requiring either synchronous communication or interference
with a current display/behavior of the client during the server
communications. Although the AJAX name refers to the use of the
eXtensible Markup Language (XML), it is also possible to use java
script object notation (JSON) to enable such asynchronous
communication.
[0066] Consequently, in the example of FIG. 3, an AJAX request 314
may be sent from the browser 310 to the REST API 306, which
initially includes the support text 110. Then, processing by the
NLP service 308 may result in identification of various sentence
parts and named entities within the support text 110, so that
annotated text 316 may be transmitted back to the browser 310 which
includes identification of the discovered sentence parts and named
entities. For example, as described herein, the original support
text 110 may be modified within the browser 310 to highlight or
otherwise indicate portions of the support text 110 which includes
the annotated text 316 (e.g., the identified sentence parts and
named entities).
[0067] In this way, for example, a user of the support client 304,
such as an IT support provider, may easily visualize portions of
the support text 110 used to characterize a necessary support
record type and associated mapping between portions of the support
text 110 and individual data fields thereof. Consequently, the user
may indicate agreement or disagreement with results included within
the annotated text 316, so that the annotated text 316 may be
corrected or otherwise updated accordingly.
[0068] FIG. 4 is a flowchart 400 illustrating more detailed example
operations of the system 100 of FIG. 1 and the system 300 of FIG.
3. In the example of FIG. 4, some or all of the sentence part score
calculator 134, the named entity score calculator 136, and the
support record type identifier 138 may be trained (402), e.g.,
using the trainer 130 and associated training set 132.
[0069] The support text 110 may then be received by way of a single
field dialog, such as the support dialog box 108 of FIG. 1 (404).
As described, the use of such a single field dialog allows an IT
support provider or other user to enter support details, including
support details conveyed during a current, direct interaction
between the IT support provider and the customer requiring support,
in a smooth, natural manner, without interruption associated with
moving between different screen items. This approach helps ensure
that core items of relevant information that may be required to
generate support tickets (e.g., including IT incidents, work
orders, customer support tickets, or requests for change) can be
gathered from the natural conversation stream between the IT
support provider and the customer, entered as the support text 110
within the single field of the support dialog box 108. As such,
forms containing one or more sets of individual input dialogs
corresponding to core parts of a support record are not required.
Moreover, as referenced, the IT support provider is not required to
move between multiple dialogs, which is disruptive to the natural
conversational flow but which would be required otherwise to gather
necessary, relevant information. As a result, the IT support
provider may be required to use fewer keystrokes, pointer moves,
and such or mouse clicks, or other physical interactions with the
support client 104, resulting in a faster, more convenient
experience for both the IT support provider and the customer.
[0070] During the time that the IT support provider or other user
is entering the support text 110, and/or after direct submission of
the support text 110 from the customer requiring support, the
support text 110 may be communicated to the support server 102/302
using AJAX, and thus to the REST API 306 of FIG. 3, as illustrated
by the AJAX request 314. In this way, the NLP service 308 of FIG.
3, parameterized to assist in implementing or otherwise providing
the text analyzer 112 of FIG. 1, may receive the support text
110.
[0071] In operation, the NLP service 308 of the text analyzer 112
may proceed to identify sentences within the support text 110
(406). For example, such processing may include segmentation of the
support text 110 for such identification of distinct sentences
within the support text 110. For example, such segmentation may be
implemented based on predefined boundary characters, such as period
(".") symbols, or other punctuation marks, spaces, characters, or
other symbols.
[0072] Subsequently, a list of words (including abbreviations and
individual numbers) may be generated from the identifying sentences
(408). For example, the NLP service 308 implementing the sentence
parser 118 of FIG. 1 may execute a tokenization process to generate
such a list of words from the discovered sentences. In this
context, white spaces and other unneeded or undesired portions of
the support text may be removed.
[0073] Sentence parts may then be identified, including associated
mapping performed with respect to sentence part tables (410). For
example, the sentence part identifier 120 associated with the NLP
service 308 may perform a tagging operation in which an
identification of sentence context is used to identify included
sentence parts.
[0074] For example, word type, in the sense of part of grammar, may
be determined, such as whether a word is a noun, adjective, or
verb. Word context, such as whether various words should be
considered singly or in combination, may be considered. Moreover,
context in this sense should be understood to relate to the type of
support being provided, and may include support-related entities
such as, e.g., proper nouns, telephone numbers, dates and times,
and specialist entities related to the business or activity for
which support is being provided. In the latter example, such
specialist entities for IT support may include, e.g., common
technical descriptors, such as IP addresses, as referenced above.
Further, such context may be understood to be imparted through the
use of the referenced sentence part tables, in which
support-specific terms or phrases may be included, so that specific
words from the support text 110 may easily be classified as
relating to specific support-relevant concepts. Specific examples
for operations associated with such sentence part identification
are described herein, e.g., with respect to FIG. 5, below.
[0075] The support text may be interpreted in real-time or near
real-time, using language analysis of the entered text. In some
implementations, embedded indicator symbols may be used, such as,
e.g., a specific character marking the fact that the following text
will be a reference to a customer's name. Such indicator symbols
may be made available to the IT support provider utilizing the
system 100/300, e.g., in conjunction with the support dialog box
108 in the browser 310. In this way, key pieces of information
relevant to generation of the support data records 116 may be
identified quickly, easily, and accurately.
[0076] Named entities within the identified words of the support
text 110 may be identified, including executing queries against
support-related data 128 (412). For example, the named entity
extractor 122 may utilize the query generator 124 to execute such
queries against the support-related data 128, as described herein.
More particularly, the named entity extractor 122 may perform
identification operations to characterize each entity isolated by
the sentence part identifier 120 (or the sentence parser 118),
using a sequence of automated research actions, including, e.g.,
generic dictionary searches, database queries of a broader support
system, or other query techniques. Specific examples for
identifying such named entities are provided below, e.g., with
respect to FIG. 6 and associated discussion thereof.
[0077] As may be appreciated, identified entities will include
those relevant to the support process, including names of people,
identifiable assets and objects having representative entries in
support system databases (or external databases), or other
support-relevant entities. Further examples of categories for
identification would include specific computer devices, employees,
knowledge articles, or historical data related to past support
incidents.
[0078] By executing such queries in real-time or near real-time, as
portions of the support text 110 are received, the text analyzer
112 and associated NLP service 308 may intelligently identify
most-relevant supporting resources, which may then be provided, as
referenced, in conjunction with the support dialog box 108 and the
browser 310. For example, such information may be provided in a
selectable list, for easy selection and use thereof by the IT
support provider at the support client 104/304. Consequently, a
requirement to instigate a search manually may be mitigated or
obviated, and the IT support provider may be ensured of being made
aware of available, useful resources, with a minimum effort on the
part of the IT support provider.
[0079] Then, sentence parts and named entities identified within
the support text 110 may be highlighted, and related data and
actions may be displayed (414), e.g., within the support dialog box
108 of FIG. 1 and/or the browser 310 of FIG. 3. For example, as
referenced above, such visual highlighting within the support text
110 serves to identify relevant portions of the support text 110 to
the IT support provider or other user. Consequently, subsequent
actions may be triggered by clicking on, or otherwise selecting,
the highlighted text.
[0080] In the example of FIG. 4, weightings for each sentence part
and named entity may be provided, relevant to potential support
types (416). For example, the sentence part score calculator 134
and the named entity score calculator 136 may assign such
weightings as scores for the various sentence parts and named
entities. Examples of such weightings are provided below with
respect to FIGS. 5 and 6.
[0081] The various weightings may then be aggregated or otherwise
combined to determine at least one support type to be associated
with the support text (418). For example, the support record type
identifier 138 may execute such aggregation operations with respect
to the calculated scores received from the score calculators 134,
136. Examples of such aggregation operations are also provided
below in the context of FIGS. 5 and 6.
[0082] In this way, at least one support type may be determined to
be associated with the support text 110, whereupon feedback on the
determined support type may be obtained (420). For example, as
described above with respect to FIG. 1, the score modifier 140 may
be utilized to interact with the IT support provider at the support
client 104 to make any necessary corrections or additions with
respect to the calculations and other operations provided by the
score calculators 134, 136, and the support record type identifier
138. Specific examples of such score modifications are described
and illustrated below with respect to FIGS. 7 and 8.
[0083] Finally in the example of FIG. 4, an individual support
record may be generated, including the filling of individual data
fields of the support record using the various identified sentence
parts and named entities (422). For example, the support record
field mapper 142 may map appropriate ones of the sentence parts and
named entities to a specific support record of the support type
identified by the support record type identifier 138.
[0084] The determination of support record type and associated,
individual support data record may also be facilitated through the
use of a confirmation process executed using interactions with the
IT support provider at the support client 104/304. For example, an
example of the presentation and confirmation operations for
confirming a determined support record type is provided below with
respect to FIG. 9, while an example of specific mapping operations
used to fill individual support record data fields is provided
below, with respect to FIG. 10.
[0085] FIG. 5 is a table 500 illustrating example operations of the
sentence part score calculator 134 of FIG. 1. As shown, the table
500 includes a column 501 of identified sentence parts obtained
from the support text 110. Meanwhile, columns 502, 504, and 506
illustrate example support record types for various types of
support that may be associated with the various identified sentence
parts of the column 501.
[0086] More particularly, as illustrated, the column 501 includes a
row 508 in which the word "needs" has been associated by the
sentence part identifier 120 with a person. That is, the identified
sentence part "needs" has been determined as being classified as an
expression of a need of a human being, as opposed, for example, to
an expression of a need of a piece of hardware for additional
hardware or software.
[0087] In this context, the row 508 further indicates relative
likelihoods within the columns 502, 504, 506 that the identified
sentence part of the column 501 relates to a corresponding type of
support record. Specifically, as shown, the column 502 includes a
support record of a type related to a "service request," while the
column 504 relates to a support record of the type "incident," and
the column 506 relates to a support record of the type "change
request." As may also be observed from the table 500, the relative
likelihoods of the columns 502, 504, 506 (i.e., 0.80, 0.15, and
0.05, respectively) add to 100%, and thus represents all possible
support record types that may be associated with the identified
sentence part of the column 501, in the example of FIG. 5.
[0088] Similar comments apply to the rows 510, 512, and 514. As
shown, each of these rows includes an identified sentence part that
is also associated with a type "person," in the same sentence as
the row 508. As shown, the row 510 includes an identified sentence
part "would like," the row 512 includes an identified sentence part
"cannot," and the row 514 includes the identified sentence part
"can't." Again, each of the rows 510-514 include corresponding
relative likelihoods that the included identified sentence part
should be associated with a given support record type specified by
the columns 502-506.
[0089] Further in FIG. 5, rows 516 and 518 illustrate further
examples in which identified sentence parts of the column 501 are
associated with a different type than the type "person." As shown,
the sentence part type illustrated in the rows 516, 518 is the type
"verb," so that the included identified sentence parts are
understood to be classified as generic parts of speech or grammar.
Then, as with the rows 508-514, relative likelihoods of being
associated with the support record types of the columns 502-506 are
included in corresponding columns.
[0090] As may be observed, the same identified sentence part may be
classified according to two or more types. For example, as shown,
the identified sentence part "can't" is identified as both a person
and a verb in rows 514, 516, respectively. In the example, the
corresponding relative likelihoods of being associated with the
support record type of the columns 502-506 are identical in the
illustrated example, but could be different for the same word,
depending on such differing classifications thereof.
[0091] Thus, FIG. 5 illustrates the sentence part score calculator
134 may be configured to analyze sentence parts identified by the
sentence part identifier 120 and the associated, parameterized NLP
service 308. Predefined, configurable mapping tables may be used,
for each input language, to provide a weighting for each identified
sentence part, based on its likelihood of appearing as part of a
particular type of support record, as illustrated and described
above.
[0092] Then, the support record type identifier 138 may proceed to
use the provided weightings for each sentence part to determine,
specifically, a most likely type of support record. For purposes of
providing a simplified example, the weightings may be applied on a
simple summed basis.
[0093] Then, in the context of such a simplified example, a
sentence of the support text 110 may include the phrase "Bob Smith
can't open . . . ," which may then be analyzed in accordance with
the techniques just described above with respect to FIG. 5 to
obtain the following sentence parts and associated weightings:
[0094] Parts of sentence:
[0095] {type=person; value="Bob Smith"} (unweighted)
[0096] {type=verb; value="can't"} (Service Request weighting=0.15,
Incident weighting=0.80, Change Request weighting=0.5)
[0097] {type=verb; value="open"} (Service Request weighting=0.34,
Incident weighting=0.33, Change Request weighting=0.33)
[0098] Then, a probability of the original sentence fragment
corresponding to a particular type of support record may be
expressed as: .SIGMA.(request type weightings for
each)/.SIGMA.(overall weightings). For the example, this expression
would then evaluate to a "service request" of likelihood of
(0.15+0.34)/2.0=0.24, to an "incident" likelihood of
(0.80+0.33)/2.0=0.57, and a "request for change" likelihood of
(0.05+0.33)/2.0=0.19.
[0099] Thus, in the simplified example, and purely with respect to
the identified sentence parts of the table 500 of FIG. 5, the most
probable record type for the particular example text fragment would
be "incident." Of course, similar analysis could be performed for
the entirety of the support text 110. Moreover, it will be
appreciated that such operation of the support record type
identifier 138 have been described with respect to FIG. 5 only with
respect to the sentence part score calculated by the sentence part
score calculator 134. As described, such analysis may be
supplemented with similar analyses performed by the support record
type identifier 138 with respect to an output of the named entity
score calculator 136.
[0100] Specifically, for example, FIG. 6 is a table 600
illustrating operations of the named entity score calculator 136.
Analogously to the table 500, the table 600 includes a column 601
which includes identified name entities, along with columns 602-606
that correspond to specific types of support records. As may be
observed, for the sake of simplicity, the column 602, 604, 606
correspond directly to the columns 502, 504, 506 of table 5.
[0101] As shown, a row 608 may thus include a named entity of
"customers assigned a laptop," while the row 610 may include the
named entity "data sentence server," and the row 612 includes the
named entity "non-IT employee as customer." As described above, the
classification of each such named entity as such may be facilitated
through operations of the query generator 124 in querying
support-related data 128, and resulting named entity types may
ultimately be scored based on associated lookup tables which
provide a likelihood that each named entity would appear in a
corresponding support record type.
[0102] Although not specifically illustrated with respect to the
example of the table 600 of FIG. 6, it may be appreciated that the
type of summation operations described above with respect to FIG. 5
may similarly be executed in the context of the table 600 of FIG.
6. That is, the support record type identifier 138 may perform a
summation, or other appropriate type of aggregation operation, with
respect to the contents of the columns 602-606, in order to
determine a most probable type of support record for the group of
named entities as a whole. Still further, additional summation or
other aggregation operations may be performed to combine the
results of such summation operations in the context of FIG. 6 with
corresponding results obtained in the context of FIG. 5. In this
way, as described in detail below with respect to FIG. 9, a final
determination of a most likely type of support record may be made
for a specific type of support text.
[0103] FIGS. 7 and 8 are tables 700 and 800, respectively,
illustrating example operations of the score modifier 140 of FIG.
1. As already referenced above with respect to the score modifier
140, such operations generally relate to potential adjustments made
to the determined probability or likelihood of each sentence part
or named entity corresponding to a particular support record type.
Such adjustments maybe made, for example, based on previous
accuracy of probabilistic results, administrator-configured
adjustments, and/or on other applied adjustment factors.
[0104] For example, one example adjustment may include
cumulatively-applied adjustment factors, incremented and
decremented when a user indicates that an existing support record
type identification was incorrect. In such cases, future
determinations for each term used for the incorrect classification
would be subject to an absolute or proportional adjustment. In some
implementations, such adjustment factors may be stored as separate
objects, thereby allowing them to be reapplied in situations in
which the underlying weight/scores are changed (which may occur,
for example, during application of a new set of vendor-supplied
data).
[0105] In the example above, and with reference to FIG. 7, a simple
percentage increment may be used as an example adjustment factor,
so that FIGS. 7 and 8 illustrate an illustrative adjustment
process. In the example, a customer's design laptop has been
detected as a named entity by the named entity extractor 122, and
the support record type identifier 138 has determined that the most
likely or correct support record type to be applied is "incident."
In fact, the IT support provider may indicate that the actual,
correct support record type should have been "request." Then, a
percentage adjustment factor may be stored and applied for future
computations.
[0106] Thus, in the table 700, a column 701 identifies identified
named entities. Column 702-706 identify potential support record
types, as already described with respect to corresponding columns
502-506 of FIGS. 5 and 602-606 of FIG. 6. Then, a column 708 may be
included for the provision of a percentage adjustment factor for
the service request likelihood of columns 702. Similarly, a column
710 may include an incident adjustment factor corresponding to the
incident weighting of the column 704, and the column 712 may
include a change request adjustment factor corresponding to the
change request weighting of the column 706.
[0107] Then, in the example just referenced, a row 714 may include
a named entity of "customers assigned laptop." The corresponding
likelihood in column 702-706 thus corresponds to the likelihoods of
the row 608 of FIG. 6. As shown, the column 708 includes a 1%
adjustment factor for the service request weighting of the column
702, while the column 710 indicates a negative 1% adjustment factor
for the incident weighting of the column 704. Finally in the
example of the table 700, no adjustment factor is included in the
row 714 for the column 712.
[0108] As referenced, and illustrated with respect to FIG. 8, the
various adjustment factors of the column 708-712 may be adjusted
incrementally, e.g., in conjunction with each determination of
support record type made by the support record type identifier 138.
For example, during a subsequent receipt of support text, a
determination of "incident" may again be made incorrectly, and
thereafter corrected to "change request." Then, as shown in cell
802 of the table 800, the incident adjustment factor of the column
710 may be incremented from -1% to -2%, while the adjustment factor
for "change request" in column 712 is illustrated as being
incremented to the value of 1% in the cell 804.
[0109] Thus, FIG. 7 illustrates and example in which a
determination of "incident" is corrected to "service request," and
a corresponding adjustment factor is applied. After receipt of
subsequent support text, a determination of "incident" is again
changed, but this time to "change request." As shown in the example
of FIGS. 7, 8, such scenarios may result in the iterative,
incremental changes illustrated, so that, over time, an accuracy of
the support record type identifier 138 may be improved.
[0110] FIG. 9 is a table 900 illustrating operations of the support
record type identifier 138 in presenting and confirming an
identified support record type. That is, as described, the support
record type identifier 138 is configured to execute a summation or
other type of aggregation operation, based on scores received from
the score calculators 134, 136, to thereby obtain one or more most
likely types of support record to be generated (subject to
modifications made and tracked in conjunction with the score
modifier 140, as just described).
[0111] In the example of FIG. 9, the table 900 illustrates an
example technique for presenting such selections of support record
types to the IT support provider or other user of the system 100,
for confirmation or modification thereof. Of course, the
illustrated presentation method is intended merely as a
non-limiting example, and actual presentation methods may vary
according to preferences of the system designer or
administrator.
[0112] In the example of FIG. 9, the table 900 includes a column
901 representing a computed probability of most likely record type,
while the column 902 illustrates corresponding actions that may be
taken in conjunction therewith. As shown, a row 904 indicates that
if a computer probability is 70%-100%, the action taken may include
an automatic selection of the associated support record type, with
retrospective correction allowed. If the computed probability of
the most likely support record type falls between 50%-70%, as shown
in the row 906, then the user may be presented with a preselected
option for confirming the determined, most likely record type,
along with an option to implement one quick confirmation thereof or
to select from a list of other possible support record types,
ordered by probability. Finally in the example of FIG. 9, as shown
in the row 908, if the computer probability falls under 50%, then
the user may simply be presented with an open selection of
potential support record types, again ordered by probability.
[0113] Of course, again, FIG. 9 is intended merely as an example.
The various thresholds of the rows 904-908 may thus be adjusted in
accordance with a preference of an administrator, and the various
associated actions may be similarly modified or adjusted in any
desired, suitable fashion.
[0114] FIG. 10 is a table 1000 illustrating example operations of
the support record field mapper 142 of FIG. 1. That is, after
selection of the appropriate support record type, as just described
with respect to FIG. 9, the support record field mapper 142 may
proceed to populate the support record in question, i.e., may
proceed to fill individual data fields thereof using identified
sentence parts and named entities, in accordance with configurable
or hard-coded mappings.
[0115] In the example of FIG. 10, a set of mappings that may be
used in conjunction with the above examples may be executed with
respect to one or more support record types, and for the named
entity "customer's assigned laptop." Of course, such mappings may
vary depending on the type of support record type selected.
[0116] Specifically, as shown in the table 1000, a column 1001
corresponds to an identified record type, while a column 1002
specifies required mappings for the identified named entity
"customer's assigned laptop." Then, in the example, in a row 1004,
a support record type of "incident" is illustrated as corresponding
to mappings associated with related configuration items (asset
record). In the row 1006, a support record type of "request" is
illustrated as being associated with mapping to a field "tag
number" corresponding to a tag number of the asset in question
(i.e., the customer's laptop). Finally in FIG. 10, a row 1008
indicates that the support record type "request for change" may not
be associated with any currently available mapping.
[0117] Thus, FIGS. 1-10 and associated description illustrate
features and functions which enable an IT support provider or other
user to obtain information regarding the potential IT support issue
in a natural, conversational manner. In such context, a support
record type may be determined during the course of a conversation
between the IT support provider and a customer or other user
requiring assistance, or immediately thereafter. Consequently, it
is unlikely to be necessary to defer decisions on record types to
be created, or to experience a need to correct a previously-created
support record that was generated incorrectly. Moreover, support
data may be entered in correct data fields of support data records
in the first instance, minimizing the chance for error or omission
of necessary data.
[0118] Further, the IT support provider is enabled to collect
specific, core pieces of support data required to build each
support record. Such items of information may be defined, for
example, by operational requirements of individuals and teams
dealing with the various types of support record on behalf of the
customer (and may also be documented in industry frameworks, such
as the IT infrastructure library (ITIL)). Thus, the IT support
provider obtains necessary information quickly, with minimal
requirements for numbers of different interface components
used.
[0119] Further, in scenarios in which an IT support provider
receives support text directly from the customers, such as may
occur when using email interfaces, IT support providers need not
retrospectively inspect each such submission after the fact to
determine its correct support record type and break out individual
pieces of required information therefore. Instead, the systems and
methods of FIGS. 1-10 may perform such inspections, determinations,
and extractions, and may thereafter input required support data
into the various, multiple data fields of the selected support
record of the corresponding support record type.
[0120] Implementations of the various techniques described herein
may be implemented in digital electronic circuitry, or in computer
hardware, firmware, software, or in combinations of them.
Implementations may implemented as a computer program product,
i.e., a computer program tangibly embodied in a non-transitory
information carrier, e.g., in a machine-readable storage device
(computer-readable medium) for processing by, or to control the
operation of, data processing apparatus, e.g., a programmable
processor, a computer, or multiple computers. A computer program,
such as the computer program(s) described above, can be written in
any form of programming language, including compiled or interpreted
languages, and can be deployed in any form, including as a
stand-alone program or as a module, component, subroutine, or other
unit suitable for use in a computing environment. A computer
program can be deployed to be processed on one computer or on
multiple computers at one site or distributed across multiple sites
and interconnected by a communication network.
[0121] Method steps may be performed by one or more programmable
processors executing a computer program to perform functions by
operating on input data and generating output. Method steps also
may be performed by, and an apparatus may be implemented as,
special purpose logic circuitry, e.g., an FPGA (field programmable
gate array) or an ASIC (application-specific integrated
circuit).
[0122] Processors suitable for the processing of a computer program
include, by way of example, both general and special purpose
microprocessors, and any one or more processors of any kind of
digital computer. Generally, a processor will receive instructions
and data from a read-only memory or a random access memory or both.
Elements of a computer may include at least one processor for
executing instructions and one or more memory devices for storing
instructions and data. Generally, a computer also may include, or
be operatively coupled to receive data from or transfer data to, or
both, one or more mass storage devices for storing data, e.g.,
magnetic, magneto-optical disks, or optical disks. Information
carriers suitable for embodying computer program instructions and
data include all forms of non-volatile memory, including by way of
example semiconductor memory devices, e.g., EPROM, EEPROM, and
flash memory devices; magnetic disks, e.g., internal hard disks or
removable disks; magneto-optical disks; and CD-ROM and DVD-ROM
disks. The processor and the memory may be supplemented by, or
incorporated in special purpose logic circuitry.
[0123] To provide for interaction with a user, implementations may
be implemented on a computer having a display device, e.g., a
cathode ray tube (CRT) or liquid crystal display (LCD) monitor, for
displaying information to the user and a keyboard and a pointing
device, e.g., a mouse or a trackball, by which the user can provide
input to the computer. Other kinds of devices can be used to
provide for interaction with a user as well; for example, feedback
provided to the user can be any form of sensory feedback, e.g.,
visual feedback, auditory feedback, or tactile feedback; and input
from the user can be received in any form, including acoustic,
speech, or tactile input.
[0124] Implementations may be implemented in a computing system
that includes a back-end component, e.g., as a data server, or that
includes a middleware component, e.g., an application server, or
that includes a front-end component, e.g., a client computer having
a graphical user interface or a Web browser through which a user
can interact with an implementation, or any combination of such
back-end, middleware, or front-end components. Components may be
interconnected by any form or medium of digital data communication,
e.g., a communication network. Examples of communication networks
include a local area network (LAN) and a wide area network (WAN),
e.g., the Internet.
[0125] While certain features of the described implementations have
been illustrated as described herein, many modifications,
substitutions, changes and equivalents will now occur to those
skilled in the art. It is, therefore, to be understood that the
appended claims are intended to cover all such modifications and
changes as fall within the scope of the embodiments. It should be
understood that they have been presented by way of example only,
not limitation, and various changes in form and details may be
made. Any portion of the apparatus and/or methods described herein
may be combined in any combination, except mutually exclusive
combinations. The embodiments described herein can include various
combinations and/or sub-combinations of the functions, components
and/or features of the different embodiments described.
* * * * *